I'm an applied ML researcher focused on AI safety, evaluation, and building systems that work reliably at scale. I spent 10+ years building ML systems at Hinge, Apple, and Google, and I'm now focused on technical AI safety research and engineering.
Most recently at Hinge, I built LLM safety evaluation pipelines and ran bias and fairness analysis on the core matching algorithm. At Apple, I built podcast search and recommendation systems and toxic content classifiers serving millions of users. At Google, I built ML evaluation frameworks for AI-generated content. I recently completed BlueDot Impact's Technical AI Safety program, studying evaluation methodology, deception detection, adversarial robustness, and interpretability.
I also make things with code, clay, and synthesizers. You can find my creative projects here.
Reach me at rickymendel at gmail dot com.